1. Introduction
Surge-type glaciers are typically characterized by short periods of rapid flow (active phase), interspersed with extended periods of slow or near-stagnant flow (quiescence) (e.g. Benn and Evans, Reference Benn and Evans2010). During the quiescent phase, low ice velocities result in ice mass build up in the reservoir zone, often located in the glacier's accumulation area, while substantial mass is lost in the receiving zone down-glacier, usually located in the ablation area. When a surge occurs, ice mass is rapidly redistributed from the reservoir to the receiving zone, often resulting in thinning in the reservoir zone, thickening in the receiving zone and an advance of the glacier front in some cases (e.g. Benn and Evans, Reference Benn and Evans2010; Jiskoot, Reference Jiskoot, Singh, Singh and Haritashya2011). Surge-type glaciers comprise only ~1% of glaciers globally, and are primarily concentrated in specific regions including Alaska-Yukon, where at least 322 surge-type glaciers have been identified through direct observation of physical changes such as increased velocity and rapid terminus advance, or due to the occurrence of indicative features such as looped surface moraines (Post, Reference Post1969; Clarke and others, Reference Clarke, Schmok, Ommanney and Collins1986; Clarke and Holdsworth, Reference Clarke, Holdsworth, Williams and Ferrigno2002; Sevestre and Benn, Reference Sevestre and Benn2015).
Theories of surge initiation in temperate glaciers have focused on hydrologic triggers based on transformation of the subglacial drainage system from channelized to distributed (Kamb and others, Reference Kamb1985; Fowler, Reference Fowler1987; Kamb, Reference Kamb1987; Murray and others, Reference Murray, Strozzi, Luckman, Jiskoot and Chrisakos2003). While the duration of surge events and their recurrence interval are unique to each glacier, surges of temperate glaciers typically last for ~1–2 years, with an intervening quiescent phase of a decade or longer (e.g. Eisen and others, Reference Eisen, Harrison and Raymond2001, Reference Eisen2005; Bevington and Copland, Reference Bevington and Copland2014; Sevestre and Benn, Reference Sevestre and Benn2015; Kochtitzky and others, Reference Kochtitzky2019). However, growing evidence suggests that glacier flow instabilities exist with a spectrum of characteristics and on variable timescales, making a simple definition of surging challenging (e.g. Herreid and Truffer, Reference Herreid and Truffer2016; Truffer and others, Reference Truffer, Haeberli and Whiteman2021). For example, the term ‘mini-surges’ has been used to describe short-term velocity pulses over a period of hours, followed by a gradual decline in velocity throughout the following day, which occur prior to a glacier surge after the reservoir has refilled with sufficient mass (e.g. Variegated Glacier; Harrison and others, Reference Harrison, Raymond and MacKeith1986; Kamb and Engelhardt, Reference Kamb and Engelhardt1987). Sund and others (Reference Sund, Eiken, Hagen and Kääb2009) also describe partial or ‘invisible’ surges, whereby there is a major mass movement in an upper basin, which does not result in a surge of the main trunk of a glacier, and does not result in a terminus advance.
Recently, studies have attempted to reconcile existing surging theories into an over-arching enthalpy theory, which suggests that surging exists where enthalpy gains (e.g. from balance velocities) cannot be effectively balanced by enthalpy losses (e.g. from heat conduction and meltwater discharge) (Benn and others, Reference Benn, Fowler, Hewitt and Sevestre2019; Benn and others, Reference Benn, Hewitt and Luckman2023). However, there are still variations in surging behaviour that are not fully explained by this theory, and further observations and analyses are required to understand surging behaviour.
It is useful to investigate surge-type glaciers to advance our understanding of dynamic ice-flow instabilities, and to better disentangle glacier-climate responses from internal variability (e.g. Yde and Paasche, Reference Yde and Paasche2010; Flowers and others, Reference Flowers, Roux, Pimentel and Schoof2011). Frequently, there are limited records which encompass the evolution of glacier surges over both quiescent and active phases. Partial surges are in particular poorly documented, and therefore this study expands our understanding of the spectrum of surging behaviours. Additionally, on a local scale, glacier surges can be hazardous to communities and infrastructure due to their rapid mass redistribution, increased calving and potential for the formation of ice dams and associated downstream flooding (Harrison and others, Reference Harrison, Shroder, Haeberli and Whiteman2015; Painter and others, Reference Painter, Copland, Dow, Kochtitzky and Medrzycka2023). For example, Kochtitzky and others (Reference Kochtitzky, Copland, Painter and Dow2020) report how repeated surges of Dań Zhùr (Donjek) Glacier, located ~30 km downstream of the terminus of Little Kluane Glacier, have blocked the Dań Zhùr River and led to the formation of several ice-dammed lakes and associated glacial lake outburst floods. While overall a limited impact, oral histories attest to glacier outburst floods having had fatal consequences for members of local Indigenous communities in the past (Cruikshank, Reference Cruikshank2001, Reference Cruikshank2005).
Here we provide the first report of a 2013–18 surge of ‘Little Kluane Glacier’, Yukon, to expand upon knowledge of glacier surges and other ice-flow instabilities in the St. Elias Mountains. We describe the surge from beginning to end to quantify the initiation location and subsequent progression, using a variety of datasets including digital elevation models (DEMs), temporally dense records of ice velocity and observations of changes in surface characteristics and terminus location. These observations are placed in the context of the behaviour of Little Kluane Glacier since the late 1940s, including the likelihood of a previous partial surge.
2. Study area
Located on the eastern side of the St. Elias Mountains in the traditional territory of Kluane and White River First Nations (Fig. 1), Little Kluane Glacier (60°45’N, 139°08’W) is ~20 km long and up to ~1.5 km wide, and ranges in surface elevation from ~1400 to 3500 m a.s.l. (Morin and others, Reference Morin, Flowers, Nolan, Brinkerhoff and Berthier2023). The glacier was previously classified as a tributary of Kluane Glacier in the Randolph Glacier Inventory (RGI 6.0) (RGI Consortium, 2017), but from recent satellite imagery and historical air photos it is clear that these glaciers have not been connected since at least 1963. No official name exists for the glacier, but historical air photo surveys completed by Austin Post in 1963 and 1965 identified it as Little Kluane Glacier, so we use that name here.
Figure 1. Little Kluane Glacier showing main locations referred to in the text. Red dashed box indicates region of interest in Figures 3 and 7. Note that the northern tributary disconnected from Little Kluane Glacier in the 1990s. The modified RGI 6.0 outline is based on the glacier's post-surge position in 2019. Base image: Sentinel-2, 03-08-2019, UTM Zone 7N. Inset: Regional map of Kluane National Park and Reserve within the St. Elias Mountains, with location of Little Kluane Glacier highlighted.
Little Kluane Glacier is land-terminating, with a section of dead ice in front of the terminus. There are extensive looped moraines on the glacier surface, in particular on the northern side of the main trunk in the ablation area. There is limited debris cover on the glacier outside of the dead ice zone and moraines. In historical air photos, and until the 1990s, a northern tributary was connected to Little Kluane Glacier (Fig. 1), but since the early 2000s it has retreated significantly. The uppermost section of the glacier is divided into three main arms (here referred to as north, upper and south), all of which have several smaller cirque tributaries (Fig. 1). The upper and south arms converge ~14 km from the terminus. Ice-penetrating radar measurements at the confluence of the north arm and main trunk in July 2021 show a maximum ice thickness of ~173 m (Fig. S1). While no direct meteorological measurements exist within the glacier catchment, the climate on the eastern side of the St. Elias Mountains is generally characterized as arctic and continental, with prevailing low winter temperatures (Newman and others, Reference Newman, Clark, Wood and Arnold2020), although since 1970 temperatures have increased at all elevations between 2000 and 6000 m a.s.l. (Williamson and others, Reference Williamson2020).
3. Methods
3.1 Data sources
To capture long-term changes in glacier geometry and surface features, historical air photos (1947–77), and optical satellite images from Landsat, Sentinel-2, RapidEye and PlanetScope (1972–2022) were obtained (Tables S1 and S2). The Royal Canadian Air Force (RCAF) acquired nadir photos during reconnaissance flights in 1947, 1951, 1972 and 1977, with the originals held in the National Air Photo Library, Ottawa; copies for this study were obtained from the University of Ottawa Map Library (https://www.uottawa.ca/library/collections/find-format/geospatial-data-maps) and the Yukon's Department of Energy, Mines and Resources Library (https://yukon.ca/en/science-and-natural-resources/research-and-monitoring/find-yukon-maps-energy-mines-and-resources) in Whitehorse. In addition, Austin Post flew a number of air photo surveys that captured Little Kluane Glacier in 1963 and 1965; these oblique images were obtained from the University of Washington Special Collections (https://www.lib.washington.edu/specialcollections/).
Landsat imagery was obtained from the United States Geological Survey Earth Explorer (https://earthexplorer.usgs.gov/) and downloaded in Level 1 GeoTIFF format. Sentinel-2 imagery was obtained from the ESA Copernicus Open Access hub and downloaded in Level-1C format (https://scihub.copernicus.eu/dhus/#/home). Images were preferentially selected from summer months under cloud-free conditions, with Landsat 1–3 (60 m resolution) and Landsat 5 (30 m resolution) imagery obtained in ~5-year intervals from 1972 to 2000, although suitable imagery during this time is sparse. Through an Education and Research Agreement with Planet Labs, RapidEye imagery (5 m resolution) was acquired from 2012 to 2019, along with PlanetScope Level 3B Analytic MS imagery (3 m resolution) for summer 2018 (Planet Team, 2017). Some of the Planet imagery required further georectification and were aligned with a Sentinel-2 base image acquired on 30 August 2021, using distinctive bedrock and topographic features as tie points.
Glacier surface elevation changes before, during and after the 2013–18 surge were characterized using DEMs from a variety of sources. DEMs were generated from SETSM WorldView imagery and ASTER Level 1A reconstructed unprocessed instrument data using the MMASTER software package (Girod and others, Reference Girod, Nuth, Kääb, McNabb and Galland2017; Table S3). Co-registration to the ASTER GDEM version 3 was completed using overlapping bedrock elevations (Shean and others, Reference Shean2016). The produced ASTER DEMs have a 10 m vertical and 2 m horizontal uncertainty (Girod and others, Reference Girod, Nuth, Kääb, McNabb and Galland2017). The SPOT DEM used in this study is 30 m resolution, and available in L1A Level (radiometric equalization, no geometric correction) from the SPOT World Heritage Data Search (https://regards.cnes.fr/user/swh/modules/60), and has an estimated uncertainty of 6 m (Korona and others, Reference Korona, Berthier, Bernard, Rémy and Thouvenot2009). A 2 m-resolution DEM produced from WorldView stereo pairs was obtained from the University of Minnesota Polar Geospatial Center (PGC), and has 3 m vertical accuracy (Noh and Howat, Reference Noh and Howat2015). All DEMs were further co-registered to each other using the method outlined by Nuth and Kääb (Reference Nuth and Kääb2011).
Glacier velocities were used to characterize behaviour prior to and after the 2013–18 surge. NASA's Inter-Mission Time Series of Land Velocity and Elevation (ITS_LIVE) program provides surface velocity mosaics derived from Landsat 4, 5, 7 and 8 imagery from 1985–2018, with a resolution of 240 m, created via the auto-RIFT feature tracking processing chain explained by Gardner and others (Reference Gardner2019a) (Table S4). Velocity magnitude in m a−1 was extracted along the centrelines of Little Kluane Glacier (Fig. 1). The ITS_LIVE velocity dataset provides the error product v_error (error in velocity magnitude in m a−1). Errors associated with the ITS_LIVE dataset, including surface skipping and sensor biases, are presented in the Regional Glacier and Ice Sheet Surface Velocities Known Issues documentation (Gardner and others, Reference Gardner, Fahnestock and Scambos2019b). To calculate errors, the error values within the modified Little Kluane Glacier RGI v6.0 outline were extracted and averaged to provide an overall value (Table S4). The ITS_LIVE point comparison tool (https://itslive-dashboard.labs.nsidc.org/) was used to download velocities at specific locations along the glacier's surface (Fig. 1). The separation time was limited to between 0 and 200 d to focus on short-term velocity peaks. Winter (January–May) RADARSAT-2 ultra-fine image pairs were acquired by Parks Canada to generate surface velocity maps (Table S5). The majority of the image pairs have 24 d separation (RADARSAT-2 repeat orbit interval), although a few sets have 48 d separation due to acquisition conflicts.
Multiple field visits were made to Little Kluane Glacier between 2018 and 2022 to observe surface conditions and the evolution of the glacier during and after the 2013–18 surge. This included visual validation of the terminus and moraine positions identified in satellite imagery, ground-based surveys to measure ice thickness and visual observations of the changes in crevassing, supraglacial hydrology and surface elevation over time. Ice thicknesses along a transect (Fig. 1, pink triangle; Fig. S1) were obtained in a ground-based common-offset radar survey on 18 July 2021, using an IceRadar impulse radar system from Blue System Integration (BSI) Ltd. Due to heavy crevassing caused by the 2013–18 surge, the survey was limited to a horizontal distance of ~263 m. The original system described by Mingo and Flowers (Reference Mingo and Flowers2010) has been updated and used in multiple studies across the region (e.g. Wilson and others, Reference Wilson, Flowers and Mingo2013; Bigelow and others, Reference Bigelow2020; Morin and others, Reference Morin, Flowers, Nolan, Brinkerhoff and Berthier2023). The transmitter pulse amplitude is ±650 V with a 2 ns rise time and pulse repeat frequency of 512 Hz. Transmitter and receiver were connected to resistively loaded dipole antennas of 5 MHz centre frequency in ice. A time window of 8000 ns was used, corresponding to an ice depth of ~670 m, more than three times the maximum ice thickness along the transect. On-board stacking was used to reduce random noise while automatically sampling a full stack every 1–2 s. De-trending, low-pass filtering and gain control were applied before manually picking data with the BSI IceRadar Analyzer software. An assumed ice velocity of 1.68 × 108 m s−1 was used to transform two-way travel time into ice depth (Bogorodskiĭ and others, Reference Bogorodskiĭ, Bentley and Gudmandsen1985). Standard error analysis of this system suggests an ice thickness precision of ±4 m, while an integrated single-frequency Garmin NMEA GPS18x has a reported horizontal accuracy of <3 m (Mingo and Flowers, Reference Mingo and Flowers2010).
3.2 Terminus length and position
Terminus positions were mapped from 1947 to 2021 using a combination of historical air photos (1947–77) and optical satellite imagery (1972–2021) (Tables S1 and S2). These nadir photos were georectified to the same 30 August 2021, Sentinel-2 base image as was undertaken for the Planet imagery, using the same methodology. A minimum of 20 tie points were used to align each of the historical photos, although more were used if the image covered a larger area. Based on a comparison of fixed landmarks with cloud-free, snow-free Sentinel-2 imagery, the estimated georectification error averaged 13 m.
Terminus positions were traced manually, and the distance from the glacier terminus to a fixed reference line at the valley mouth close to Kluane Glacier was measured (Fig. 1). The maximum terminus extent was measured in each image, rather than an average across the glacier's width, as the terminus is narrow and typically well defined. However, there were limiting factors in the terminus position mapping, including the relatively low resolution of early imagery (1972–2000), shadowing from surrounding mountains, and extensive debris cover over the lowermost part of the glacier. To maintain consistency the same person manually digitized all termini outlines, with uncertainties determined using the method described by Paul and others (Reference Paul2013). It is assumed that outline errors are equal to at least one pixel, resulting in a range in uncertainty from 3 to 60 m, with an average of 16.5 m, as imagery is heavily skewed towards newer observations from higher resolution imagery.
3.3 Supraglacial hydrology and surface features
The surface morphology of Little Kluane Glacier, including crevassing, supraglacial hydrology and moraine positions, was characterized using the georectified nadir air photos and optical satellite imagery from 1951 to 2020 (Tables S1 and S2). These were used to identify and manually measure the length and area of temporally consistent supraglacial hydrologic features such as surface channels and lakes, and how supraglacial drainage patterns evolved before, during and after a surge. The supraglacial lakes were outlined in RapidEye imagery in 2016 using the same process as the terminus positions, and area uncertainty was assumed to be as high as 5%, ranging from 276 to 1628 m2, as determined using the Paul and others (Reference Paul2013) method (Table S6).
The location and evolution of a series of looped moraines along the surface of Little Kluane Glacier were mapped in ~10-year intervals during quiescent periods, and as often as monthly during surge periods when suitable imagery was available. During quiescence, moraine positions were mapped from imagery taken between May and September, as snow often obscures the features at other times of the year. However, during surge periods it was possible to also map changes in moraines during the winter months. Changes in moraine position were quantified in relation to a set point up-glacier of the initial position of each moraine loop. Uncertainties in moraine locations are assumed to be within one pixel in the imagery used to map their position, following the methodology used for the terminus outlines (Paul and others, Reference Paul2013), equating to 10–60 m for Landsat scenes (corresponding to the quiescent periods) and 3–10 m for Sentinel-2 and Planet imagery (coinciding with surge periods).
3.4 Digital elevation model (DEM) analysis
To compute surface elevation change between 2010 and 2019, the earliest SPOT DEM (03-09-2010) was subtracted from the most recent post-surge ASTER DEM (04-08-2019) to generate a DEM of Difference. As these DEMs have different sources, we co-registered them and compared their values over stable terrain to determine their uncertainty. Based on the largest uncertainty value of the two sources, this equated to an average of ±10 m. Glacier surface elevation change was extracted at 10 m intervals along south and north arm centrelines (Fig. 1), and smoothed using a 300 m moving window to reduce noise. Mountains surrounding the glacier are steep, leading to data gaps and high comparative errors in these regions, although relative patterns on the glacier surface are clear. Changes in surface slope during the most recent surge were examined using a SPOT DEM (03-09-2010), and WorldView DEM (17-07-2016) (Table S3). Ice rim heights were estimated by extracting and analysing elevation points from the 22-08-2020 ASTER DEM.
3.5 Winter velocity mapping
Knowledge of winter velocities was required as glacier surges often initiate in the late winter or early spring in the study region (Abe and Furuya, Reference Abe and Furuya2015). These were derived from SAR imagery as feature-tracking velocities computed from optical imagery, such as those provided through ITS_LIVE, are unreliable over short periods in the winter when visible features on the glacier surface are obscured by snow, shadow and/or low light levels. SAR imagery also delivers accumulation-area velocities, which can be difficult to determine from optical imagery. Speckle-tracking was performed on winter (January–May) RADARSAT-2 ultra-fine image pairs to generate surface velocity maps (Table S5). GAMMA software was used to process the SAR image pairs, as it provides velocity results similar to other speckle-tracking methods previously applied in northern and western Canada (Waechter and others, Reference Waechter, Copland and Herdes2015; Van Wychen and others, Reference Van Wychen2018). Based on co-registered backscatter intensity images, a patch intensity cross-correlation was used to determine displacement in the azimuth and range directions from the offset between corresponding pixel blocks in each image pair (Lu and Veci, Reference Lu and Veci2016).
Velocities (in m a−1) were filtered to remove noise and mismatches. Coarse filtering was used to retain values between 5 and 5000 m a−1, which represented the lower detection limit and maximum velocities previously measured in the St. Elias Mountains, respectively (Waechter and others, Reference Waechter, Copland and Herdes2015). A custom Matlab script was used to automatically filter the data based on flow direction, where values that differed by >50° from the mean direction of neighbouring cells were discarded (Main and others, Reference Main2022). Where multiple image pair results from the same winter existed, overlapping values were averaged to produce a single velocity value, and Inverse Distance Weighted (IDW) interpolation was used to fill minor gaps (<250 m). The results were clipped to a modified Randolph Glacier Inventory version 6.0 outline, with Little Kluane Glacier separated from Kluane Glacier. Following a similar methodology to that used by Waechter and others (Reference Waechter, Copland and Herdes2015), velocities were extracted at 50 m horizontal intervals along centrelines (Fig. 1) and smoothed using a 100 m moving window.
Variations in the satellite orbital model, poor image co-registration, DEM/cross-correlation errors and layover or foreshortening effects in SAR imagery can create errors in speckle-tracking results (Schellenberger and others, Reference Schellenberger, Van Wychen, Copland, Kääb and Gray2016; Van Wychen and others, Reference Van Wychen2018). To quantify these errors, the apparent motion over stable ground was determined outside of a buffer of 100 m around the modified RGI v6.0 outline. The apparent motion is the average speed of all stable ground pixels, irrespective of x or y direction. We assume that this represents the uncertainty, with values ranging between 3 and 6 m a−1 for the 2014–20 RADARSAT-2 data (Table S5).
4. Results
Results from this study are presented below both chronologically and by dataset, from the earliest available results to the most recent.
4.1 Terminus position
The oldest available imagery of Little Kluane Glacier, a RCAF photo from 1947, indicates that the glacier terminus once pushed into Kluane Glacier, displacing the adjacent ice, and creating a slight bulge (Fig. 2a). This photograph only includes the lowermost section of the terminus of Little Kluane Glacier, with the bulge appearing to contain mainly dead ice as demonstrated by significant debris cover, potholes in the surface, and a lake and river near the valley mouth. This suggests that a surge of Little Kluane Glacier occurred some years before the image was acquired.
Figure 2. (a) RCAF air photo from 24-07-1947, demonstrating how Little Kluane Glacier previously pushed into Kluane Glacier (image ID A11014-275). Red dashed line indicates the separation between Little Kluane and Kluane glaciers. (b) Terminus positions of Little Kluane Glacier from 1947-2015. Projection: UTM 7N. Base image: Sentinel-2, 24-09-2016. (c) Change in relative terminus position from 1947-2021 in relation to distance from valley mouth. Timing of partial surge (possible initiation range) and recent full surge are highlighted in grey. (d) Terminus advance between 2015 and 2019, during recent surge. Base image: Sentinel-2, 30-08-2021.
From 1947 to 1963, the terminus retreated by up to ~2000 m, but this was not constant across the terminus due to its irregular shape in 1947 (Fig. 2a). From 1963 to 1998, the terminus retreated gradually, with a total retreat of ~980 ± 60 m (~28 ± 1.7 m a−1) (Figs 2b, c). From 1998 until 2013, the rate of terminus retreat increased significantly, with a total loss of ~1650 ± 30 m (~110 ± 2 m a−1). The terminus advanced by ~340 ± 15 m (~170 ± 7.5 m a−1) between 2015 and mid-August 2017 then remained stable until the end of May 2018, when thickening at the terminus began. After this, the terminus advanced by an additional 1400 ± 15 m (~12 m d−1), stagnating by September 2018 (Figs 2c, d). Following surge termination in 2018 the terminus position was ~2900 ± 15 m short of its maximum position in 1947. From 2018 to 2022 the terminus remained mostly stagnant, with minor variations (within 70 m) and evidence of downwasting at the end of the terminus, although debris cover and shadowing made it difficult to detect small-scale changes.
4.2 Surface features
There were distinct changes in the crevassing and supraglacial hydrology of Little Kluane Glacier over the study period (Fig. S2). From 1977 until 2012, the surface was relatively smooth, with a ~1 km long supraglacial channel (Fig. S2a, b) and small supraglacial lakes present on the surface of the north arm in the accumulation zone (Figs 3a, b). This supraglacial channel remained in approximately the same position over several decades, as it was visible in the 3 August 1977 air photo, as well as in later imagery up to 2013.
Figure 3. (a) Supraglacial lake locations on the surface of Little Kluane Glacier on 29-06-2016, during the active surge. (b) Location of water-filled crevasses and supraglacial lake, on 27-07-2018. (c) One day later on 28-07-2018, showing partial drainage of these features. Satellite images courtesy of Planet Labs. General location is indicated by red dashed box on Figure 1.
In 2013, crevasses were present over a small area of the upper north arm, as was the supraglacial channel (Fig. S2c). By 2015, significant crevassing had appeared (Fig. S2d) in an area where, in 2012, the surface was smooth (Fig. S2b). There was a progressive increase in the intensity of crevassing from 2015 to 2016 (Figs S2d, e), and by mid-June 2016 the supraglacial channel had been interrupted near its head by new, larger crevasses. Extensive crevassing covered nearly the entire surface of the upper north arm (i.e. accumulation zone) to the connection with the main glacier trunk by June 2016. Crevassing in the south and upper arms was delayed by approximately a year compared to the north arm: in 2013–14 there was limited crevassing in the south and upper arms, followed by an increase in intensity from 2016 to 2017, and finally a peak in crevassing intensity and extent by September 2018. Crevassing in the main glacier trunk began to increase in intensity between February and June 2018, was extensive by July 2018 (Fig. 3) and peaked in September 2018.
Starting in June 2016 two supraglacial lakes (Fig. 3a) formed below the north arm-main trunk connection, in an area where no lakes were observed in prior imagery. The lakes reached a maximum size of ~32600 ± 1630 m2 on 18 June 2016, and ~17400 ± 860 m2 on 29 June 2016, and had almost completely drained by late July that year. Heavy cloud cover in May and June 2017 made it difficult to monitor the lakes during this period, and only the west lake was observed to redevelop. By July 2017, the glacier morphology had significantly changed, and the east lake did not develop. From mid-July until mid-August 2017, the west lake decreased in area, although due to its small size the uncertainty estimates are too high to produce reliable estimates of its area. By September 2017, the west lake had completely drained. In 2018, the lakes again formed in approximately the same positions, but were surrounded by significant crevassing and additional ice from the north arm (Fig. 3b). RapidEye imagery from 27 to 28 July 2018 show that the east lake drained entirely within a single day (Figs 3b, c).
There were four teardrop-shaped moraines visible in the 1951 air photos, which can be interpreted as the product of past surges (Fig. 4a) (Meier and Post, Reference Meier and Post1969; Young and others, Reference Young, Flowers, Jiskoot and Gibson2022): these features were located on the north side of the glacier, were relatively evenly spaced apart and had similar structural appearances. Between 1972 and 1977, a new looped moraine (E) was created (Figs 4b, S3). The oldest looped moraine, ‘A’, was visible in the record until 1994 (Fig. 4c), when it reached the terminus. The three moraine loops (C, D and E) moved a maximum of ~1 km over the 43-year period from 1972 to 2015 (Fig. 4e), with higher velocities up-glacier and an average of 25 m a−1.
Figure 4. Looped moraine positions (labelled with capital letters) for each decade at Little Kluane Glacier in historical air photos (a–b) and satellite imagery (c–d) (listed in Tables S1 and S2); (a) 08 July 1951; (b) 11 August 1972; (c) 16 August 1994; (d) during surge initiation, 29 July 2016. Air photos are courtesy of the National Air Photo Library, Ottawa. (e) Looped moraine movement from 1972–2021. Moraines A and B are not included as they disappear before the recent surge.
After 2015, a new moraine loop (F) was formed via the transfer of ice from the north arm to the main glacier trunk between October 2015 and February 2016 (Figs 4d and S3). Due to poor image quality during the winter it is difficult to pin-point exact timing, but movement of the new bulge (moraine F) continued until September 2016 (Fig. S3f), and then slowed for most of 2017 (Fig. S3g). The largest displacement occurred between January 2018 and June 2018 (Figs S3h–i), slowing in August 2018 to background velocities through 2019 (Figs S3j–l). Existing moraine loops C, D and E were each transported ~1.7 km down-glacier during this period (Fig. 4e). Loop B (the second oldest moraine visible beginning in 1951) was no longer visible on the glacier after 2019.
4.3 Surface elevation changes
Changes in the surface elevation of Little Kluane Glacier between 2001 and 2019 indicate a transfer of mass from the top of the north arm and the main glacier trunk (over an area of ~27.5 km2) into the terminus region, resulting in an overall thickening of ~100–170 ± 10 m in the lower 2.5 km of the glacier over an area of ~7.3 km2 (Fig. 5). There was a corresponding thinning of up to ~90 ± 10 m in the upper north arm basin (Fig. 5), which is evident from a stranded ice rim that remained attached to valley walls (Fig. S4). In contrast, the upper arm and the uppermost region of the south arm (12–17.5 km up-glacier) remained relatively stable, with the main mass movement occurring from the main glacier trunk (~15–11 km from terminus) (Fig. 5b). Thinning of the south arm was more limited in area and intensity, with losses up to 70 ± 10 m along the south arm centreline (Fig. 5b). There is also evidence of localized thickening where the valley narrows, such as at the confluence between the north arm and the main glacier trunk.
Figure 5. (a) Elevation change across Little Kluane Glacier between 2010 and 2019, with overlain modified RGI 6.0 glacier outline and 2018 terminus extent. Projection: UTM 7N. Base image: Sentinel-2, 3 August 2019. (b) Little Kluane surface elevation change derived from a DEM of Difference between 2010 and 2019; dashed blue, green and yellow lines indicate where the respective arm connects to the main glacier trunk. Dashed red line indicates the location of the Dynamic Balance Line (DBL).
Between 2010 and 2016, there were localized changes to the glacier surface slope. Specifically, the slope in the confluence region of the north arm with the main trunk reversed between 2010 and 2016 (Fig. 6). In 2010, the slope of line A-A’ was 4.5%, and B-B’ was 5.2%, while in 2016 the reversed slope section of the survey line (Figs 6c, d) had a slope of −2.5 and −9.3%, respectively. The reverse slope created a ~0.13 km2 basin.
Figure 6. Changes in glacier surface topography after surge initiation. General location is indicated by red dashed box on Figure 1. (a) SPOT DEM of glacier surface on 03-09-2010 prior to surge initiation. Background image: Rapid-Eye, 06-08-2013; (b) WorldView DEM of glacier surface on 17-07-2016, as mass moves from the north arm to the main trunk of Little Kluane Glacier, demonstrating a reverse slope and resulting depression; (c) ice elevation profiles of survey line A-A’ in 2010 and 2016; (d) ice elevation profiles of survey line B-B’ in 2010 and 2016. Satellite images courtesy of Planet Labs. The relationship between ice elevation and supraglacial lake location is only shown for the west lake, as DEM coverage does not extend to the east lake.
4.4 Surface velocities
Annual average velocities obtained from the ITS_LIVE dataset (Fig. 7) revealed three main dynamic phases between 1985 and 2013. First, there was a higher velocity period from 1985 to 1996, where velocities in the main trunk ranged from ~40 to 70 m a−1, with lower (<25 m a−1) velocities in the north arm. This was followed by a lower velocity period from 1997 to 2012, when velocities across the glacier centreline averaged 10 m a−1, and never exceeded 50 m a−1. Finally, during the third period from 2013 to 2018 maximum annual velocities progressively increased, beginning at ~40 m a−1 in 2013, ~70 m a−1 in 2014, ~100 m a−1 in 2015, ~260 m a−1 in 2016, ~375 m a−1 in 2017 and finally ~830 m a−1 in 2018 (Figs 7 and S5). Velocities increased first in the north arm, followed by the main glacier trunk (Fig. S5a). Both the upper and south arms of the glacier retained relatively stable velocity patterns compared to the north arm and main trunk, with modest increases occurring mainly between 2016 and 2018.
Figure 7. Annual average ITS_LIVE velocities along the north arm centreline of Little Kluane Glacier from 1985 to 2018 (see Fig. 1 for location). Dashed black lines indicate where the north arm joins the main glacier trunk. Dashed red line indicates location of the Dynamic Balance Line (DBL), which signifies the transition between the reservoir and receiving zones.
Point measurements from ITS_LIVE image pairs (Fig. 1) indicate increasing velocities from 2016 through 2018 (Fig. S5). Mean velocity of points located in the north arm (Fig. S5a) showed steady increases through 2016, followed by slight decreases in 2017, before peaking in fall 2018. Velocities in the main trunk increased progressively towards the terminus, with a point ~12 km up-glacier from the terminus (point 3) increasing in speed ~1 year prior to a point ~8 km up-glacier from the terminus (point 2), and 2 years prior to a point ~4 km up-glacier from the terminus (point 1) (Fig. S5b). Mean velocities in the main trunk were steady from 2014 to 2016 (<100 m a−1), increased in 2017 and peaked at ~3600 m a−1 near the terminus in fall 2018 (Fig. S5b).
RADARSAT-2 wintertime velocities indicate increased ice motion along the north arm centreline from 2014 to 2020 (Fig. 8). In the upper accumulation zone of the north arm, velocities increased over 4 years, from 2014 to 2017 (Fig. 8b). In the main glacier trunk, velocities remained stable (~40 m a−1) through 2014–15, then increased in 2016 and 2017, before decreasing to background levels in 2020 (Fig. 8a). The highest winter velocities (~1000 m a−1) were recorded ~5 km up-glacier from the terminus in early 2018 (Fig. 8a).
Figure 8. (a) Winter (January–May) velocities from RADARSAT-2 data along the north arm centreline of Little Kluane Glacier, 2014–2020. Red box indicates location of zoomed in section in part (b). Dashed black lines indicate where the north arm joins the main glacier trunk. Dashed red line indicates the location of the Dynamic Balance Line (DBL). (b) Zoom-in to show velocity details for the north arm of Little Kluane Glacier from 2014 to 2019. Note: there are no data from 2018.
5. Discussion
Our observations indicate that Little Kluane Glacier underwent a full surge over the period 2013–18, with substantial changes in terminus position, supraglacial hydrology, crevassing, surface elevation and ice velocities. This is the only full surge to have occurred between 1947 and 2022. There is evidence that a partial surge occurred during the 1960s–70s, and that multiple surges occurred prior to 1947, although these are poorly constrained in terms of timing and velocities. We discuss the evidence for these events chronologically below.
5.1 Pre-1947 surges
The presence of large, looped moraines on the glacier surface suggests that at least four surges occurred prior to 1947. The spacing between each looped moraine increases from the terminus through the main trunk, from ~1100 m between loops B and C, to ~2500 m between loops D and E in 2013, which may reflect a more compressive flow regime towards the terminus, or increasing time between surging intervals. The distance between the moraines has remained mostly consistent since 1972, with only a slight decrease in distance (~200 m) between moraines closer to the terminus. Observations from the 1960–70's partial surge (section 5.2) and 2013–18 full surge (section 5.3) indicate that these looped moraines form from the episodic movement of mass from the north arm into a slower moving main glacier trunk (Fig. S3). A similar but increasing distance occurs between each loop, suggesting regularly occurring surges, although the timing between them is not well constrained. From 1989 to 2010, moraine loop B flowed down-glacier at an average rate of ~35 m a−1, loop C at ~15 m a−1 and loop D at ~25 m a−1. If we take the average of this motion of 25 m a−1, and use a contemporary glacier length of 17 km, this means that at least four partial or full surges must have occurred in the ~70 years prior to 1947 to produce them, assuming that mass balance and velocity conditions have remained relatively constant. When compared with only one observed partial surge and one full surge from 1947 to 2022, this implies that Little Kluane Glacier has experienced a decline in surging frequency, likely indicative of a slower filling of the reservoir zone and higher snowlines. This trend aligns with the prevailing strongly negative mass balance observed in the St. Elias Mountains (Hugonnet and others, Reference Hugonnet2021).
Studies from other glaciers have shown that potholes begin to form on the ice surface as early as 4 years after surge termination (Sturm, Reference Sturm1987; Partington, Reference Partington2023). If we assume that the last full surge prior to 1947 behaved similarly to the 2013–18 full surge, then by comparing the development of the surge bulge with the well-developed pothole and surface channels covering the dead ice zone at the glacier terminus in 1947 (Fig. 2a), we suggest that the pre-1947 surge likely occurred within the prior decade. This means that there was a period of at least 70 years between the two most recent full surges. In Figure 2a, the terminus extends ~2.5 km further than the most recent advance; this might have been caused by a greater pre-surge extent, a larger pre-1947 surge or a combination of both scenarios.
5.2 1960–70's Partial surge of Little Kluane Glacier
Although no full surge occurred between 1947 and 2013, there is still evidence of dynamic variability during this time. Historical air photos demonstrate that the north arm pushed into the main trunk of Little Kluane Glacier between 1963 and 1972 (Fig. S3), although the terminus did not advance during this time, instead following the long-term pattern of >4 km retreat between 1947 and 2013 (Fig. 2d). This mass movement created moraine E (Figs 4b, S3c-d), although it is smaller and more elongated compared to the other moraine loops, such as loop D.
Partial surges are described by Sund and others (Reference Sund, Eiken, Hagen and Kääb2009) as a phenomenon when a mass movement occurs but does not instigate a terminus advance. In terms of their proposed surge stages, this refers to when a surge does not develop past stage 2, where there is surface lowering in major parts of the reservoir zone and thickening in the receiving zone, but there is no pronounced acceleration across most of the glacier surface, nor is there a terminus advance. Sund and others (Reference Sund, Eiken, Hagen and Kääb2009) hypothesize that surge-type glaciers may oscillate between weaker, partial surges and more developed, full surges, and that this behaviour may explain the long build-up phase of some glaciers. They argue that while small and partial mass movements from the reservoir zone may initially delay a full surge, they could also promote a large surge through the slower, more stable, and potentially enhanced buildup of ice mass.
Between 1963 and 1972, there was a mass movement originating from the north arm to the main trunk of Little Kluane Glacier that displaced the medial moraine by ~250 m (Fig. S3); however, this did not develop into a full surge by 1977, with the bulge remaining in essentially the same position and no evidence of glacier acceleration or terminus advance (Fig. S3d). Changes in glacier surface morphology, including curved or displaced medial moraines have been used to identify partial surges in Svalbard (Sund and others, Reference Sund, Eiken, Hagen and Kääb2009, Reference Sund, Lauknes and Eiken2014) and the Pamirs (Kotlyakov and others, Reference Kotlyakov, Osipova and Tsvetkov2008; Shangguan and others, Reference Shangguan2016). We therefore interpret this as being a partial surge of Little Kluane Glacier. This event formed a new moraine loop (loop E, Figs S3c–d).
5.3 Characteristics of the 2013–18 surge
The 2013–18 surge of Little Kluane Glacier was driven primarily by the transfer of a large mass of ice from the north arm that prompted a surge in the main trunk. The extensive photographic, velocity and elevation records for this glacier provide insights into surge-related processes, including the interplay between surface topography and hydrology. Increased velocities in 2013 provided the first signal of the impending surge, followed by intensifying crevassing through 2015 and 2016, which interrupted a prominent supraglacial channel near the head of the north arm (Fig. S2). As velocities and strain rates increase at the start of a surge, the larger stress results in intense crevassing, enhancing distributed water access to the glacier bed that may further increase velocities due to the reduction in basal friction, thereby establishing a positive feedback (Dunse and others, Reference Dunse2015; Gong and others, Reference Gong2018). For Little Kluane Glacier, this increased crevassing appears to have caused the supraglacial hydrological network to change from a concentrated to a distributed system (Fig. S2), which over time may have influenced the englacial and subglacial hydrologic networks, as surface water was no longer entering the glacier interior and bed from a single point (Shepherd and others, Reference Shepherd2009), but instead through multiple locations (Colgan and others, Reference Colgan2011). As there is a delay between the increased velocities beginning in 2013 and the mass movement in 2016, we hypothesize that the evolution of the surficial hydrological network resulted in an increasing volume of water at the glacier bed, thereby increasing basal sliding, and enhancing the movement of mass from the north arm to the main trunk of the glacier.
At Little Kluane Glacier, changes in surface slope and supraglacial hydrology at the junction between the north arm and main trunk may also play a role in defining whether a partial surge of the north arm develops into something closer to a glacier-wide surge. Glacier velocities in the main trunk began to increase noticeably in 2015, a year after velocities increased in the north arm (Figs 7 and S5). Beginning in early winter 2016, movement of a large ice mass occurred from the north arm to the main trunk. The addition of this mass changed the surface topography of the main trunk above the confluence with the north arm, and created a reverse surface slope, allowing surface water to pool and form supraglacial lakes in 2016, which reformed and drained throughout 2017 and 2018 (Figs 3, 6). The following winter (2018), the glacier entered the fast, active phase (Fig. S5b), first increasing in velocity near the confluence with the north arm (Fig. S5b, point 3), with the surge then propagating down-glacier. The formation and drainage of these lakes over several years, combined with incrementally increasing glacier velocities (~155 m a−2) over a 3-year period suggest that there may have been a slow, and then rapid, shift between subglacial drainage configurations, whereby significant, localized water input to the bed over multiple seasons may have eventually overwhelmed the subglacial drainage network and provoked acceleration, until an efficient drainage network was able to form and evacuate the water (Kamb and others, Reference Kamb1985). Sund and others (Reference Sund, Lauknes and Eiken2014) suggested that a tributary is able to initiate a surge in the main trunk only if the main trunk was already in stage 1 (initial surface lowering, increased velocities) or stage 2 (larger mass displacement down-glacier), and thus ready to reach the final surge stage. However, at Little Kluane Glacier, we hypothesize that the main trunk becomes topographically and hydrologically primed for the final surge stage over a period of several years prior to the secondary surge initiation. In the absence of subglacial hydrological observations, we cannot definitively show that the change in surface morphology or hydrology (e.g. formation of supraglacial lakes and water-filled crevasses) drove the main trunk into a surge, although this may be a process that could be observed during tributary-trunk surges on other glaciers (Bhambri and others, Reference Bhambri2022).
In 2018, the surge front reached the glacier terminus region, and surface velocities peaked at ~1000 m a−1 during the winter (Fig. 8) and ~2200 m a−1 near the end of summer (Fig. S5b). The terminus began to advance in May 2018, which coincided with the highest recorded point velocity of ~3600 m a−1 ~4.5 km up-glacier from the terminus (Figs 2, S5b). The terminus underwent a total advance of ~1.4 km over spring and summer 2018 (Fig. 2c), at a mean rate of 12 m d−1. Similar to other glaciers in the region including Klutlan, Variegated and Bering (Jay-Allemand and others, Reference Jay-Allemand, Gillet-Chaulet, Gagliardini and Nodet2011; Burgess and others, Reference Burgess, Forster, Larsen and Braun2012; Altena and others, Reference Altena, Scambos, Fahnestock and Kääb2019), Little Kluane displayed a winter initiation, with the highest velocities occurring for a short period during the summer, and the surge terminated abruptly towards the end of the melt season. Considering the above characteristics, along with the already-discussed evidence of subglacial hydrological processes, we suggest that Little Kluane Glacier most closely resembles surges controlled by changes in basal water pressure, at least when considering the main trunk.
After the surge event, it is possible to classify zones of the glacier based on changes in surface elevation. The dividing threshold is a 0 ± 10 m elevation change, which delineates the two zones as suggested by Pitte and others (Reference Pitte2016). We therefore consider the region >10 km up-glacier from the terminus to be the reservoir zone (Fig. 5), and the zone from ~6 km to the terminus the receiving zone. The zone of minimal net surface elevation change, known as the dynamic balance line, occurs ~6 km up-glacier from the terminus. While the surge was mainly driven by ice movement from the north arm, ice mass from the south and upper arms was pulled along as the full surge developed in the main trunk, further suggesting that the main trunk was not in a surge phase prior to input from the north arm. The minor role of the south and upper arms in the surge is demonstrated by the relatively small changes in ice surface elevation above the south arm-main trunk confluence compared to the north arm basin (Fig. 5), and the minor change in ice surface elevation in the upper arm (Figs 5a, b).
From the data presented above, we suggest that the surging behaviour of Little Kluane Glacier follows these main phases:
1. Due to a constriction at the connection between the north arm and the main trunk (a narrow valley which is quite steep, with a trough at the bed (Fig. S1)), the glacier is unable to efficiently transfer mass from the north arm to the main trunk, resulting in mass accumulation within the basin during quiescence.
2. After decades of accumulation, the ice mass in the north arm becomes thick enough that glacier velocities increase. Velocities increase over a period of 2–3 years, and crevasses begin to appear, disrupting persistent supraglacial drainage channels and routing water to the glacier bed in a distributed fashion.
3. Eventually the north arm ice mass is able to overcome the resistive forces acting upon it from the valley walls, bed and abutment with the main trunk, resulting in a large mass movement from the tributary to the main trunk. In one instance (~1963–72) this did not initiate a full surge (Fig. S3).
4. Instigation of a full surge of the glacier may be facilitated by mass transfer from the north arm altering the surface slope below the junction of the north arm and main trunk (Figs 3, 4), as occurred around 2016 (Fig. 6). A surface slope reversal allows water to pool and then drain into the glacier, possibly overwhelming the subglacial drainage system of the main trunk and initiating a surge there.
5. The surge front then propagates down-glacier, causing the terminus to advance, and driving enhanced ice motion in the south and upper arms due to a combination of longitudinal coupling with faster-flowing downstream ice, a steepening in surface slope and a reduction in backpressure where they meet the main trunk.
6. Active surging ends as ice mass reaches the lower part of the glacier, lowering the driving stress and reducing glacier velocities to ~10 m a−1. For Little Kluane Glacier, this coincided with the end of the melt season in September 2018.
7. The quiescent stage begins, and ice mass starts accumulating again in the north arm and along the main glacier trunk above the dynamic balance line.
5.4 Comparison of the 2013–18 surge of Little Kluane Glacier with other surges in the St. Elias Mountains
Surging glaciers can be difficult to characterize due to lack of high-frequency temporal data from surge initiation through termination, resulting in uncertainties concerning where surges initiate, how they progress, the length of active/quiescent cycles and whether a surge bulge is present. In Alaska-Yukon region, 322 surge-type glaciers have been previously identified (Sevestre and Benn, Reference Sevestre and Benn2015), with 151 in the St. Elias Mountains of Yukon alone (Clarke and others, Reference Clarke, Schmok, Ommanney and Collins1986). When compared to surges of some of the other glaciers in the region with comprehensive records (Dań Zhùr, Sít’ Kusá, Nàłùdäy and Variegated), the 2013–2018 surge of Little Kluane Glacier has a longer quiescent phase (~45 years compared to ~5–20 years for the others), and a long initiation phase (~5 years of initiation then ~1 year of intense surging, compared to 1–2 years of total active surging for the others). Several glaciers to the southeast, however, have quiescent phases exceeding 30 years. The most recent surges of Dusty Glacier (~80 km away) and Tweedsmuir Glacier (~140 km away) were <3 years in duration and preceded by quiescent intervals of ~36 years (Young, Reference Young2023) and ~34 years (Sharp, Reference Sharp2021), respectively. Fisher Glacier (~110 km away) had a quiescent phase of ~40 years, and an active phase of ~2.5 years, with a slow multidecadal increase in velocity between the mid-1980s and 2013 (Partington, Reference Partington2023). Several smaller polythermal glaciers in the region, including Trapridge Glacier (Clarke and Blake, Reference Clarke and Blake1991; Frappé and Clarke, Reference Frappé and Clarke2007) and ‘South Glacier’ (De Paoli and Flowers, Reference De Paoli and Flowers2009), formerly surged in a manner similar to the glaciers above with multidecadal quiescent intervals, but more recently underwent prolonged ‘slow surges’ where the high velocities associated with typical surges failed to develop.
5.5 Impact of geometric restrictions on glacier surging
Other studies (Jiskoot, Reference Jiskoot, Singh, Singh and Haritashya2011; Abe and others, Reference Abe, Furuya and Sakakibara2016; Kochtitzky and others, Reference Kochtitzky2019; Nolan and others, Reference Nolan, Kochtitzky, Enderlin, McNabb and Krutz2021) have suggested that restriction of ice flow is crucial in influencing surging behaviour of specific glaciers. The restrictions affecting glaciers can take a number of forms, including bedrock sills and overdeepenings (Flowers and others, Reference Flowers, Roux, Pimentel and Schoof2011; Jiskoot, Reference Jiskoot, Singh, Singh and Haritashya2011; Lovell and others, Reference Lovell, Carr and Stokes2018), constrictions in valley width (Kochtitzky and others, Reference Kochtitzky2019), trunk–tributary interactions (Kotlyakov and others, Reference Kotlyakov, Osipova and Tsvetkov2008; Sund and others, Reference Sund, Eiken, Hagen and Kääb2009; Jiskoot and others, Reference Jiskoot, Fox and Van Wychen2017) and icefalls (Nolan and others, Reference Nolan, Kochtitzky, Enderlin, McNabb and Krutz2021). For some glaciers, these features may act as a potential barrier to efficient ice flow due to increased friction during quiescence, and hence may encourage mass (and thus, energy/enthalpy) to build up in the reservoir zone (Sevestre and Benn, Reference Sevestre and Benn2015). For example, Eisen and others (Reference Eisen, Harrison and Raymond2001) found that cumulative mass balance played a role in the duration of the surge cycle of temperate Variegated Glacier, which was influenced by the location of a bedrock sill.
At Little Kluane Glacier, we suggest that the narrow connection between the north arm and the main glacier trunk acts as a flow constriction that, along with the backstress of the much larger main trunk, restricts ice flow, allowing ice mass to build. Eventually, the ice mass accumulates sufficiently to steepen the surface slope and overcome the frictional forces at this constriction, inducing a mass movement. The size of this ice flux (either relatively small, as in 1960s–1970s, or large, as in 2013–18) results in either a partial or a full surge, which may be dependent on several factors, including the cumulative mass balance between surges, and/or the state of the subglacial hydrological network of the main trunk.
6. Conclusions
Little Kluane Glacier experienced a progressive increase in glacier velocities in the north arm basin from 2013 to 2017, prior to the initiation of a full surge which occurred in 2018. A total of six previous dynamic instabilities were identified based on the presence of looped moraines, which includes one full and one partial surge documented here. A partial surge was identified in the mid-1960s to early-1970s, whereby mass movement was observed from the north arm tributary to the main trunk, but did not initiate a full surge and was smaller than the mass displacements that occurred during the 2013–18 surge. It appears that the narrow connection between the north arm tributary and the main trunk is important in defining the nature and extent of surges through the creation of a flow restriction. Evidence for this is provided by the initiation of the 2013–18 surge at the very top of the tributary through increased crevassing, then a mass movement from the tributary to the main trunk 2 years later, followed by a full surge in the main trunk. Additional data are needed to determine whether topographic or geometric factors are underpinning glacier surges at other locations in the St. Elias Mountains, and therefore detailed valley and bed elevation profiles are necessary to fully understand the role that geometry plays in glacier surging in this region.
This study demonstrates that unstable flow of Little Kluane Glacier depends on a number of factors which are not yet fully understood. These could include glacier geometry, whereby a partial surge may delay a full surge by dissipating a portion of the stored energy, and therefore appear to promote conditions that favour more gradual reservoir growth. Another hypothesis is that the initial mass movement from a tributary to the main glacier trunk could initiate a full surge by producing a reverse slope that establishes the necessary conditions for water to pool on the surface, which eventually drains and overwhelms the subglacial drainage network. This may act as the trigger that enables a partial surge to evolve into a full surge affecting most of the glacier area. This study underlines the importance of data frequency and resolution when describing dynamic instabilities, and highlights that there are a variety of intersecting behaviours that can occur during a surge.
Supplementary material
The supplementary material for this article can be found at https://doi.org/10.1017/jog.2024.35.
Data
The RADARSAT-2 glacier velocities generated for this study are available from the Polar Data Catalogue under CCIN reference number 13309. Original RADARSAT-2 data are available through MDA Geospatial Services (https://www.asc-csa.gc.ca/eng/satellites/radarsat2/order-contact.asp). All satellite imagery (Landsat, Sentinel-2, RapidEye and Planet), air photos and ITS_LIVE velocity products are publicly available from data sources listed in the methods.
Acknowledgements
Fieldwork for this project was conducted on Kluane and White River First Nations Traditional Territory, and the authors thank the communities and Parks Canada for permission to work there. Support for this research has been provided by the University of Ottawa Geographical, Statistical and Government Information library (particularly René Duplain, Pierre Leblanc and Hugo Crites), Canada Centre for Mapping and Earth Observation (Natural Resources Canada), Northern Scientific Training Program, Canadian Space Agency (RADARSAT Constellation Mission Data Utilization and Application Plan), Natural Sciences and Engineering Research Council of Canada, New Frontiers in Research Fund, Canada Foundation for Innovation, Canada Research Chairs Program, University of Ottawa Research Chairs Program, Ontario Research Fund, Parks Canada, Kluane Lake Research Station, Outpost Research Station and Polar Continental Shelf Program. Support for travel to conferences and attendance of courses was generously provided by the ArcticNet Network of Centres of Excellence Canada Student Training Fund, GlacioEx and RemoteEx. We thank Library and Archives Canada for providing the historical air photos. We are grateful for the thoughtful and constructive comments provided by editor R. Bhambri, and by three anonymous reviewers.
Author contributions
B. M. and L. C. designed the study. B. M. processed the surficial hydrology, crevassing and moraine datasets, completed the analysis and wrote the initial draft of the manuscript with input from L. C., G. E. F., C. F. D. and W. V. W.. G. E. F., S. S. and W. K. provided processed datasets. All authors revised the paper prior to submission.
Open access
